QPyTorch：一个低精度算术仿真框架
QPyTorch: A Low-Precision Arithmetic Simulation Framework
摘要：
低精度训练降低了计算成本，并产生了有效的模型。最近在开发新的低精度训练算法方面的研究通常依赖于模拟来经验性地评估量化的统计效果，同时避免构建特定硬件的大量开销。为了支持这一实证研究，我们引入了QPyTorch，一个低精度的算术模拟框架。QPyTorch在PyTorch中原生构建，提供了一个方便的界面，可以最大限度地减少可靠转换现有代码以研究低精度训练所需的工作。QPyTorch是通用的，并支持各种精度，数字格式和舍入选项的组合。此外，它利用一种有效的融合内核方法来减少模拟器开销，从而能够模拟大规模的现实问题。QPyTorch可在https://github.com/Tiiiger/QPyTorch上公开获取。
英文摘要：
Low-precision training reduces computational cost and produces efficient models. Recent research in developing new low-precision training algorithms often relies on simulation to empirically evaluate the statistical effects of quantization while avoiding the substantial overhead of building specific hardware. To support this empirical research, we introduce QPyTorch, a low-precision arithmetic simulation framework. Built natively in PyTorch, QPyTorch provides a convenient interface that minimizes the efforts needed to reliably convert existing codes to study low-precision training. QPyTorch is general, and supports a variety of combinations of precisions, number formats, and rounding options. Additionally, it leverages an efficient fused-kernel approach to reduce simulator overhead, which enables simulation of large-scale, realistic problems. QPyTorch is publicly available at https://github.com/Tiiiger/QPyTorch.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
